Jàng bu boole
Federated learning dafay tàggat ab model buñ bokk ci aparey wala mbootaay yu bari te duñu dajale seeni done yu bees ci benn barab.
Résumé
Only model updates travel to the server, so the sensitive data stays where it lives.
Plongeur bu xóot
Ci tàggat-yaram bu jaar yoon, dañuy boole done yépp ci serveur yu mag yi. Federated learning flips lii: ab model global lañuy yónnee ñi ci bokk (telefoon, hopitaal, bànk), ku nekk ci ñoom tàggat ci ay done boppam, te coppite yi ci diisaay bi kese lañuy yónnee. Serveur bi dafay moyenne coppite yooyu ci model global bu gëna baax ba noppi baamtu ko. Google moo dugal xalaatu Gboard, gëna mëna wax luy am ci klaweer ci ay milioŋ ciy telefon te duñu yebbi li nit ñi bind. Bokk na ci anam yi ñuy jëfandikoo ay done yu nëbbu, yuñ yamale, wala yu yaatu lool ngir ñu mëna toxal, lu melni dokimaa yu ñeel liñ faj, yu tasaaroo ci hopitaal yi. Jafe-jafe yi bokku na ci aparey yu wóorul, done yu wuute lool ci diggante ñi ci bokk (done yu amul IID), ak ni coppite yu bees yi mën nañu ba leegi sacc ay leeral, moo tax ñu boole ko ak pexe yu nëbbu.
Gis-gis xarala
Algorithm biñ gëna xam mooy Federated Averaging (FedAvg): kiliyaan bu nekk dafay def ay jéego yu bari ci wàllu gradient, ginaaw ga serwër bi dafay jël moyenne bu pondéré ci poid yu bees yi, ñu koy faral di pondéree ci limu done yi kiliyaan bu nekk am. Ndax kiliyaan yi dañuy tàggat seen bopp ci jéego yu bari balaa ñuy sync, rond yu jokkoo yi dañuy wàññeeku bu baax méngoo ak yónnee gradient bu nekk. Ngir tere yeesali done yi di sacc, sistem federaal yi dañuy yokk agrégation bu wóor, buy may serwër bi mu gis limu boole, ak privacy bu wuute, biy dugal bruit buñ kalibre.
njeextalu pexe
Njëgg ak budget
Dogal yi architecture di jël dañuy indi njariñ ak njëgu liggéey bi ay at ci ginaaw.
dogal yu gëna leer
Njàngalem xarala yi dafay jàppale ekip yi ñu tànn li gën, te baña yam ci li gëna bees daal.
Xool kalite
Tanneef yu gëna baax ci wàllu ingeñër dina wàññi jafe-jafe yi ci wàllu wóor ci liggéey bi.
Ëlëgu jàng bu boole
Njàngalem federaal mingi joge ci klawieer yi dem ci jëfandikoo mbootaay ci wàllu faju, xaalis, ak IoT, fu sàrt yu melni HIPAA ak GDPR di def ay done yu jafe. Xaarandil lëkkaloo bu gëna dëgër ak nëbbëtu bu wuute ak dajale bu wóor, boole ci kaadar yu melni TensorFlow Federated, Flower, ak NVIDIA FLARE ñuy màgget ngir liggéey. Benn frontière buy màgg mooy federaal fine-tuning ci modeli làkk yu mag, may mbootaay yi ñu boole seen doole ngir gëna suqali benn model ci mbindu kumpa. Gëstu bi gëna am solo mooy gëna mëna jëflante ak ñi bokk ci jàngat bi te kenn mënu leen wóolu.
Doxal ci àdduna dëgg
Google Gboard dafay gëna suqali kàddu yi ci topp ak emoji yi ci telefon yi te doo yebbi benn tombu.
Hopitaal yi dañuy bokk tàggat xeetu nataali diagnostic te duñu séddoo done yuñ aar ci malaad yi.
Bànk yi di liggéeyandoo ci xeeti gis-gis njuuj njaaj ci noonu lañuy tëye jëflante bu nekk ci bànk yi.
Apple dafay personaalise man-mani yi ci aparey bi lu ci melni QuickType ak digle Siri ci jëfandikoo njàngum dëkk bi.
Risk yi ak balustrade yi
Optimize benn benchmark mën na nëbb ñakk kattan yu gëna yaatu ci sistem bi.
Njëg li ñuy fay ci infrastructure yi ak ci toppatoo dañuy faral di suufeel.
Bu sistem yi di gëna xawa jafee xam, jafe-jafe yi am ci wàllu kaaraange ak seetlu mën nañu gëna bari.
Roadmap ngir samp gi
Mandargal latency, kalite, ak njëg yi laata ngay jëfandikoo.
Benchmark ci biir sargal ak done yu dëggu.
Jumtukaay bi di saytu njuumte yi, derive bi ak njeextalu jëfandikukat bi.
Waajal rollback ak yooni tontu ci jafe-jafe yi laata ngay eskale.
Weyal di banneexu
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Federated Learning quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Gis bi ci topp
CI/CD ngir jàngu masin
Laaj yi ñuy faral di laaj
What is Federated Learning?
Federated learning dafay tàggat ab model buñ bokk ci aparey wala mbootaay yu bari te duñu dajale seeni done yu bees ci benn barab. Coppite model yi kese ñooy dem ci serwër bi, kon done yu am solo yi dañuy des fi ñu dëkk.
Lan mooy xalaat bu mag bi ci njàng mu boole?
Njàngale mu federaal dafay denc done yu bees yi ci gox bi; ku nekk ci ñoom tàggat ci ay done boppam te du yónnee lenn ludul yeesali model yi serwër bi boole.
Lan la algorithm moyenne fedérée (FedAvg) di def ci serwër bi?
FedAvg dafay boole poñ yi kiliyaan yi tàggat ci gox bi ci benn xeetu àdduna bu gëna baax, ñuy poñ kiliyaan bu nekk ci limu done yi mu jëfandikoo.
Ban produit moo gëna jëfandikoo njàngum federaal ngir gëna mëna wax luy am?
Google jëfandikoo na jàngat buñ boole muy Gboard ngir gëna mëna wax luy waaja am ci ay milioŋ ciy telefon te du jël li jëfandikukat yi bind.
Lan moo waral njàngum federaal lu am solo la ci wàllu faju?
Done yu wérgi-yaram dañuy nekk ci bopp te dañu leen yamale, kon denc leen ci barab bi ba noppi tabax xeetu ñu bokk méngoo ak yoon yu melni HIPAA moo gëna baax boole ko.
Lan moo waral yeesali model yi kese duñu nekk lu nëbbu ci saasi?
Gradient yi ak coppite yi ci diisaay bi mën nañu fësal ay leeral ci misaali tàggat yaram, moo tax ñu yokk aggregation bu wóor ak privacy bu wuute.